Machine learning and eel passage = True?

The project aims to generate new knowledge about the movement patterns of silver eels and how these are affected by environmental factors, in order to design more effective and sustainable passage measures at hydroelectric power plants. Using high-resolution data and advanced analysis, the researchers want to improve understanding of eel behaviour and support both ecological and economic decisions.

The project is based on a unique dataset from the Swedish river Ätran back in 2017, where 98 silver eels were tagged with acoustic transmitters with depth sensors. The researchers followed the eels as they passed Herting’s power plant, both via a deflector with an escape opening and a natural fishway. Previous analyses have provided detailed information about water flows and other environmental factors, but how these affect the movements of eels is still unknown.

Now, using an analysis protocol for salmon smolts, researchers will study how the bottom structure, water movements, temperature and light affect the eels’ route choices and behaviour. The results can be used at other locations and help to create better solutions for eel passage at hydroelectric power plants. The project is also a first step towards predictive models that can eventually test and improve measures before they are implemented.

Background

Sustainable use of natural resources requires balancing different ecosystem services. In today’s society, the need for energy is increasing in line with growing electrification, while environmental awareness is rising and demands for renewable energy sources with low climate impact are increasing. Hydropower is one of the few large-scale renewable energy sources that can also be regulated, which makes its environmental impact and potential for optimisation highly relevant.

In Sweden, the environmental conditions for most of the country’s approximately 2,300 hydropower plants will be renewed over the next 20 years. Measures to restore water connectivity are both costly and location-dependent, which makes improved knowledge of fish passage ecologically and economically valuable.